• Please review our updated Terms and Rules here

Pdp11/34 restoration

I've found a problem, but I'm uncertain how it could cause this problem.

I connected a scope to the system, and noticed a problem with the +5V power supply. I'll figure out how to capture the waveform and post it tonight, but...

There is some ringing in the 5v supply, for lack of better words. 5v is running between 4.4 and 5.6 volts, with a waveform pattern like this ( read top to bottom )

!
!
-----------
--------
---
-
!
!

It's a poor description, but it looks to me like maybe a weak capacitor or over/under voltage regulator.

I imagine an edge sensitive signal could be messed up by something like this...
 
Nice work - this could be a significant find. I’m not certain it’s completely responsible for your first issue, but it sounds like a problem that definitely needs to be fixed. (I’m assuming those measurements were taken under load.)

Which chassis are you running (BA11-K 10 ½” or BA11-L 5 ¼”)? Compared to the H765, I would say the smaller H777 power supply used in the BA11-L is more prone to problems and trickier to troubleshoot, but there is good reference information available on bitsavers.org. Please keep us posted.
 
Yes, the scope work was done under load. The chassis is the BA11-L with the H777 power supply. I don't know much about non-digital electronics - power supplies, for instance - but the maintenance manual mentions a master clock that calibrates the voltage regulator; I'm going to check out that frequency as per the documentation.
 
It looks like a frequency of about 45KHz. It may be just +/- 0.5V of common mode noise depending on how good your return reference was on the scope. If you are familiar with scopes, try to make an A-B measurement using two probes, Channel 1 of the 5V bus and channel 2 on the 5V return. This will be more of a pseudo differential measurement and will tell the true story. Switch the scope Mode to 'ADD' and use channel 2 'INVERT' to get a 'channel 1 minus channel 2 reading. Both channels must be on the same volts/div setting. Let me know if you need more info on how to make this type of measurement.
 
Be careful of those power supplies, they are switching supplies and use a high DC voltage on the unregulated side and that can be lethal! You may also want to go and look at the “PDP-11/44 Restoration “ link being that was dealing with power supply issues although I think you may just have some week electrolytic capacitors on your system.
 
It could be the 50khz ref clock, I'll check. In any case, the H777 manual specifies a common mode noise variation of less than 2%, not the 20% I'm seeing...
 
I'm pretty sure the H777 doesn't create high voltages, the reference power supply is 25-48 VAC, I think. Anyway, I am always extremely careful around anything with 120VAC inputs.

I really hope you are right about the caps, and it wouldn't surprise me. I'm taking the power supply into a local shop tomorrow...the tech ere is described as a wizard.
 
Could someone with a 34 out there (qbus, maybe) try a simple test for me?

Halt the machine, press any key on the RS232 console an A will do, use the front panel to examine 777560, it should be 000200, press CTRL/INIT, and examine 777560 again. It should be 000000.

Thanks very much.
 
Yes, my 34 dose that. The status register (777560) has 000200 in it after receiving a character, and clears to 000000 after hitting “INIT” and I can see the octal for what the character is in 777562 but here is the fun part you can deposit an octal into register 777566 like 000063 and a 3 will appear on your con!
 
Yes , but it’s nowhere near as much fun on the 11/23 as it is on a system with a keypad because your using the con in the first place to communicate with the system. On the unibus system you’re entering numbers and pushing buttons on the CPU, and almost as if by magic as soon as you deposit a value in 777566 it there on the terminal.
Can only imagine how much fun it would be using an interface with all the switches and lights and having to enter everything in binary but have to think that using the keypad and entering things in octal may be quickest.
 
Back
Top